Skip to main content

Table 3 \(\text {NN}_{RS \& LP}\), \(\text {NN}_{SM}\) and LRs characteristics for the HFDB and the IDB

From: Serial electrocardiography to detect newly emerging or aggravating cardiac pathology: a deep-learning approach

  Architecture   HFDB IDB
[16 13 12] [11 9 1]
\(\text {NN}_{RS \& LP}\) Learning AUC (%) 99 98
Testing AUC (%) 84 83
CI (%) [73–95] [75–91]
ACC (%) 75 76
\(\text {NN}_{SM}\) Learning AUC (%) 86 77
Testing AUC (%) 83 73
CI (%) [72–94] [60–87]
ACC (%) 75 67
LR Learning AUC (%) 89 88
Testing AUC (%) 61 77
CI (%) [46–75] [68–86]
ACC (%) 54 71